This example shows Chernoff faces for lawyers' ratings of twelve judges
Chernoff faces, invented by applied mathematician, statistician, and physicist Herman Chernoff in 1973, display multivariate data in the shape of a human face. The individual parts, such as eyes, ears, mouth, and nose represent values of the variables by their shape, size, placement, and orientation. The idea behind using faces is that humans easily recognize faces and notice small changes without difficulty. Chernoff faces handle each variable differently. Because the features of the faces vary in perceived importance, the way in which variables are mapped to the features should be carefully chosen (e.g. eye size and eyebrow-slant have been found to carry significant weight).[1]
Detail
Chernoff faces themselves can be plotted on a standard X–Y graph; the faces can be positioned X–Y based on the two most important variables, and then the faces themselves represent the rest of the dimensions for each item. Edward Tufte, presenting such a diagram, says that this kind of Chernoff-face graph would "reduce well, maintaining legibility even with individual areas of 0.05 square inches as shown ... with cartoon faces and even numbers becoming data measures, we would appear to have reached the limit of graphical economy of presentation, imagination, and let it be admitted, eccentricity".[2]
Extensions
Asymmetrical faces
In 1981, Bernhard Flury and Hans Riedwyl suggested "asymmetrical" Chernoff faces;[3] since a face has vertical symmetry (around the y axis), the left side of the face is identical to the right and is basically wasted space – a point also made by Tufte.[4] One could have the 18 variables that specify the left be one set of data, but use a different set of data for the right side of the face, allowing one face to depict 35 different measurements. They present results showing that such asymmetrical faces are useful in visualizing databases of identical twins, for example, and are useful in grouping as pairs of Chernoff faces would be.[3]
Chernoff fish
Julie Rodriguez and Piotr Kaczmarek use "Chernoff fish", where various parts of a cartoon fish are used to encode different financial details.[5]
In literature
In Peter Watts' novel Blindsight (2006), a transhuman character is seen using a variant of Chernoff faces. This use is explained by the character as a more efficient method of representing data, for a large portion of the human brain is devoted to facial recognition. [6]
In the 2014 sci-fi short story "Degrees of Freedom" by Karl Schroeder, Chernoff faces make a prominent appearance as a future technology, supporting the communication of aggregate sentiment and perspective.[7]
^Edward R. Tufte, The Visual Display of Quantitative Information, p. 97: "Halves may be easier to sort (by matching the right half of an unsorted face to the left half of a sorted face) than full faces. Or else an asymmetrical full face can be used to report additional variables (Flury & Riedwyl 1981). Bilateral symmetry doubles the space consumed by the design in a graph, without adding new information."
"Facial Representation of Multivariate Data", David L. Huff, Vijay Mahajan and William C. Black. Journal of Marketing, Vol. 45, No. 4 (Autumn, 1981), pp. 53–59.Published by: American Marketing Association
"FACES-A FORTRAN Program for Generating Chernoff-Type Faces on a Line Printer", Danny W. Turner and F. Eugene Tidmore. The American Statistician, Vol. 34, No. 3 (Aug., 1980), p. 187 Published by: American Statistical Association
Gonick, L. and Smith, W. The Cartoon Guide to Statistics. New York: Harper Perennial, p. 212, 1993.
Moriarity, S. (1979). "Communicating financial information through multidimensional graphics". Journal of Accounting Research 17, 205-224.
Stock, D. and Watson, C. (1984). "Human judgment accuracy, multidimensional graphics, and humans versus models". Journal of Accounting Research 22, 192-206.
Information Visualization: Perception for Design (2004), Ware C.
NovoSpark Visualizer - an advanced visualization tool that enables qualitative analysis of multidimensional data through the exploration of a graphical image.
